Reece Hart <hart.reece@gene.com> writes:
> After a system crash, postgresql 8.1.4 restarted but reported that I
> have an apparent wraparound:
> ...
> 2006-07-13 14:03:40 PDT [10092] LOG: next MultiXactId: 5475264; next MultiXactOffset: 13765525
> ...
> 2006-07-13 14:03:40 PDT [10092] LOG: could not truncate directory "pg_multixact/offsets": apparent wraparound
> 2006-07-13 14:03:40 PDT [10092] LOG: could not truncate directory "pg_multixact/members": apparent wraparound
> There are two files left in pg_multixact:
> -rw------- 1 postgres postgres 16K 2006-07-13 14:13 pg_multixact/members/00D2
> -rw------- 1 postgres postgres 144K 2006-07-13 14:13 pg_multixact/offsets/0053
That's odd. Those files correspond to the next MultiXactId and
MultiXactOffset, so there shouldn't have been any such complaint.
[ looks at code... ] I wonder if this is happening because
shared->latest_page_number hasn't been set up yet when we do the
end-of-recovery checkpoint.
> In
> http://www.mail-archive.com/pgsql-general@postgresql.org/msg76635.html
> Tom Lane implies that I can probably ignore these messages.
No, I was saying that the "invalid server process ID -1" was harmless.
The "apparent wraparound" is a distinct issue, and I'd ask you the same
question I asked Thomas: do you continue to get those log messages
during subsequent checkpoints?
regards, tom lane